You can run AI at the edge, if your infrastructure supports it
THE REGISTER EXPLAINER Companies are increasingly running AI applications close to where
data is generated and consumed. That means everywhere from branch
offices to retail sites and industrial facilities. These use cases
often share a key characteristic: They can’t wait for a round trip to
a hyperscale region.
Drop media element here …
Why does cloud-first break at the edge?
Processing data locally cuts the cost and delay of moving high-volume
streams to the cloud while strengthening privacy and compliance, a
calculus that sharpens as regulators move in. Frameworks like the EU
AI Act demand auditabile inferencing for high-risk AI workloads.
How does edge AI computing change security?
Distributed sites expand the attack
surface, which makes low-level hardware-based security and
centralized policy control essential. HPE ProLiant edge servers embed
a silicon root of trust in the iLO management chip to block
compromised firmware. That kind of defense matters at edge locations,
where malicious actors can physically reach hardware far more easily
than at a hardened datacenter.
While competitors use off-the-shelf chips, HPE
designs its own baseboard management controller silicon for the role
— a level of protection suited to threat surfaces that extend to
the back office behind a cash register, on a manufacturing floor, or
even a back office storage room.
How can I keep my AI operating in non-datacenter
environments?
Datacenter hardware tends to fail amid
dust, temperature swings, weak power and intermittent connectivity.
The HPE ProLiant DL145 Gen11 is about half the depth of a DL365 and
quiet enough at ~55 dB to sit in an office. With a new processor,
this rugged unit supports GPU such as the NVIDIA RTX PRO™ 4500
Blackwell, tolerates temperature variations, and includes built-in
air filtration.
Can I manage edge AI at scale?
HPE Compute Ops Management helps to manage
distributed computing environments by providing global visibility
from a cloud-native console. Administrators can deploy firmware
updates, monitorhealth, and provision new edge servers. Forrester
found
that organizations using the tool spend up to 75 percent less time
managing remote servers, with substantial savings in travel and
manual effort.
Treating the edge as a business-critical platform
rather than a sprawl of isolated boxes lets AI initiatives scale
without IT overhead scaling alongside.
Whether you’re running a vision system to spot
defects on a manufacturing line or machine learning to flag operating
anomalies on oil field equipment, AI runs more efficiently as it gets
closer to the point of use. Choosing equipment that’s physically
resilient, secure and easy to manage will be the difference between
edge excellence and distributed dystopia.
Sponsored by HPE.